Levi Neuwirth
53854ed803
tests/compile_mode_acceptance.rs (34 tests): spawn shape + header + exit markers; read-only under dispatch; child-boundary stderr merge in emission order; stdin EOF; group kill/leader-exit/escalation/ ledger bites incl. the redirected TERM-ignoring survivor and the pipe-holding-descendant tick-latency bound; starter-rule parsing with 0-based normalization and severity posture; sub-1 fail-closed; severity override + malformed-rule containers; unterminated final line; RET/n-p/M-g n/M-g p/C-x ` navigation pins with the diag fallback; recompile + q-target discipline; supersede baseline; all seven undo/redo chords table-driven; M-x undo after a completed run recovering via buffer.after-edit; no-hook shrink and same-length newline-moving replace with anchor epochs; ANSI SGR/CR with rendered-cell attachment proof surviving RET-then-M-,; killed-buffer teardown; grep locations panel, kill-mid-search + masking prevention, root retention; shell-command M-!; round-trip pins. tests/compile_mode_crdt_acceptance.rs: a chord-triggered full run converges byte-identically on two replicas (mid-session generated- buffer snapshot adoption), and a synthetic accepted replica edit triggers the immediate recovery marker, converging across the causal-reorder seam. Fixes found by the suite: compile.lua's CR handling now scans the current line start from the buffer (the REPL discipline) instead of using the per-batch parse position — a same-batch CR previously let a progress line overwrite earlier output; malformed Lua patterns are rejected (and counted) at validation time via a probe match. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01VoiEyuPjoBhvwACf8HAnLB |
||
|---|---|---|
| .. | ||
| api | ||
| commands | ||
| hooks | ||
| keymaps | ||
| menus | ||
| packages/repl | ||
| runtime | ||